Intercom systems provide communication between two or more remotely located individuals. More advanced intercom systems provide conference-type communication, simultaneously connecting several intercom stations so that several individuals can communicate with groups and sub-groups of other system users.
Conventional conference-type intercom systems include a switch matrix, commonly referred to as a crosspoint switch, which allows any user to communicate with any selectable mix of the remainder of the users. Crosspoint switches are normally used instead of direct point-to-point connections between source and destination equipment for all but the smallest implementations. A thorough discussion of prior art crosspoint switching schemes can be found in U.S. Pat. No. 5,483,528 to Christensen, incorporated herein in its entirety by reference.
With crosspoint switches, all stations are connected directly to the crosspoint switch matrix which makes connections between the sources and destinations internally. In order to accommodate large numbers of switched audio channels, intercom systems typically distribute the crosspoint matrix across a number of bussed circuit modules or cards. Each module typically controls switch closures for connecting audio to or from a small group of intercom stations for which it is responsible. Each module typically employs a small local computer whose duties include, but are not limited to:                (a) Making and breaking audio crosspoint closures;        (b) Data communications with those intercom stations connected to it;        (c) Data communications with other matrix modules via a central data communications controller.        
A significant disadvantage of crosspoint switches is the geometrically increasing size and cost of the switch matrix relative to the number of intercom users to be interconnected. For example, although a four station matrix requires only 16 switches and an interconnect backplane containing 4 circuit paths, a 100 station matrix requires 10,000 switches and dedication of 100 backplane circuit paths. Thus the available matrix-of-switches crosspoint topology is practical only for relatively small systems.
Noise considerations also become more important as matrix-switch-based intercom systems grow in size. Resistive summing of multiple analog sources into a common destination results in signal-to-noise degradation. Driving multiple destinations from a single source requires the addition of analog buffers to meet impedance-matching fanout requirements. These buffers also degrade signal-to-noise performance.
Partially addressing these problems, the Christensen patent discloses a system which interconnects intercom stations, making use of Time Division Multiplexing (TDM). In this system, multiple audio signal sources (stations) are periodically sampled and digitized at a high frequency and interleaved cyclically onto a parallel bus. At each destination (station) on the bus, a receiver picks from the data stream the digitized samples associated with a given source or combination of sources. It converts these signals to analog voltages and reconstructs the original analog waveform or combination of waveforms by interpolating (integrating) between the sequential samples taken from the bus.
Although TDM greatly improves the efficiency of audio transport in intercom systems, traditional intercom systems (even the system disclosed in Christensen) suffer limitations in the transport of the large volume of control data upon which intercom system operation depends. Consider, for example, the requirement for any practical intercom system, that an audio connection be established very quickly, and usually bi-directionally. Control messages must pass between intercom stations over narrow bandwidth data channels, separate from the audio channel, typically through an intermediary or central data communications controller to establish the audio connection. The station initiating the conversation must notify the second station, through the intermediary, that it desires to establish a talk-listen connection. The second station then acknowledges the first station's request and replies with a request to establish its own talk-listen connection, also through the intermediary processor. As system complexity and size increase, the number of messages passed between stations escalates, and the limited bandwidth of the data links, especially when implemented as serial data channels, tends to become a bottleneck for data flow. If the intermediary processor and its communications links delay the control message-passing to any great degree, the delay can become objectionable to users of the intercom. The party attempting to initiate a conversation may begin to speak before the link is established, causing the second party to receive a truncated, unintelligible message. The common intercom topology employing serial control communications through an intermediary processor, even including TDM-based audio signal transport, can be overwhelmed by the volume of data transactions when an intercom system reaches a critical size.
Also, if the central communications processor malfunctions, the capacity for data communication is lost and cannot be reestablished until the central processor is repaired or replaced. Avoiding this failure mode by using redundant central communications processors tends to be awkward and complicated.
An additional flaw in some available systems can be found in the use of the TDM bus for transport of mixed and attenuated audio signals. This redundant use of the TDM bus narrows the available bandwidth for more critical primary audio transport between stations and would preferably be avoided.
Therefore, as demand grows for ever larger intercom systems, the industry needs a mechanism which provides wider bandwidth data communications to support the increased channel capacity of TDM audio transports. Preferably such a system would provide immunity from catastrophic single-point failure without the appreciable complexity of redundant central data communications processors.